“Yandex” neuronets can now search for information in the archives of “Soviet sport” for 100 years

Short description

Summarize this content to 100 words Yandex’s artificial intelligence recognized 24,000 issues of the newspaper “Soviet Sport” over a hundred years, from 1922 to 2022. Now they are available in the “Archive Search” service: users will be able not only to view scans of newspaper pages, but also to search for information by various numbers and articles. According to “Yandex”, the neural network was adapted to work with newspaper pages and their special layout: A2 format spreads with columns, insets, advertisements with small letters on a specific paper of low density, from which it is more difficult to read information. The company’s specialists manually processed thousands of pages, highlighting on each tens and hundreds of text blocks from headings and columns to chess notations and inscriptions on athletes’ uniforms. In the future, the updated recognition technology will help to better navigate other types of complex layout, for example, when working with handwritten archives.As for the archives of “Soviet Sport”, in them you can find, for example, an interview with the legend of Soviet football Igor Netto, an article about an unusual approach to transfers in the middle of the last century in Brazil, and a note about the fact that the Austrian Football Association is no longer intends to tolerate players with “Beatles-style” hair in the squads. “Yandex” also made a small selection of funny articles:• The team was not allowed to participate in the regatta because of the driver, stoker and sailor (1922);• At the competition, one of the discs on the bar turned out to be heavier than the other (1951);• The athlete did not leave the house on the advice of astrologers and broke his hip in the bathroom (1963);• Collection of fake games of the USSR Chess Championship (1977).The project was implemented as part of Yandex’s partnership with “Soviet Sport” and the National Electronic Library (NEB), which collects documents from the funds of Russian libraries and provides them to users in digital form.The company launched “Archive Search” at the beginning of 2023. The service helps to quickly find references to people, settlements and events in handwritten documents of the 18th-20th centuries deciphered by neural networks.In March, Yandex announced that its neural networks had deciphered 300,000 pages of metric books from the Central State Archive of the Moscow Region. Now, in addition to issues of “Soviet sport”, the service database contains more than 7 million pages of historical documents from the archives of Moscow, Moscow, Irkutsk, Orenburg and Novgorod regions, as well as several municipal archives.

“Yandex” neuronets can now search for information in the archives of “Soviet sport” for 100 years

Yandex’s artificial intelligence recognized 24,000 issues of the newspaper “Soviet Sport” over a hundred years, from 1922 to 2022. Now they are available in the “Archive Search” service: users will be able not only to view scans of newspaper pages, but also to search for information by various numbers and articles.

According to “Yandex”, the neural network was adapted to work with newspaper pages and their special layout: A2 format spreads with columns, insets, advertisements with small letters on a specific paper of low density, from which it is more difficult to read information. The company’s specialists manually processed thousands of pages, highlighting on each tens and hundreds of text blocks from headings and columns to chess notations and inscriptions on athletes’ uniforms. In the future, the updated recognition technology will help to better navigate other types of complex layout, for example, when working with handwritten archives.

As for the archives of “Soviet Sport”, in them you can find, for example, an interview with the legend of Soviet football Igor Netto, an article about an unusual approach to transfers in the middle of the last century in Brazil, and a note about the fact that the Austrian Football Association is no longer intends to tolerate players with “Beatles-style” hair in the squads. “Yandex” also made a small selection of funny articles:

• The team was not allowed to participate in the regatta because of the driver, stoker and sailor (1922);

• At the competition, one of the discs on the bar turned out to be heavier than the other (1951);

• The athlete did not leave the house on the advice of astrologers and broke his hip in the bathroom (1963);

• Collection of fake games of the USSR Chess Championship (1977).

The project was implemented as part of Yandex’s partnership with “Soviet Sport” and the National Electronic Library (NEB), which collects documents from the funds of Russian libraries and provides them to users in digital form.

The company launched “Archive Search” at the beginning of 2023. The service helps to quickly find references to people, settlements and events in handwritten documents of the 18th-20th centuries deciphered by neural networks.

In March, Yandex announced that its neural networks had deciphered 300,000 pages of metric books from the Central State Archive of the Moscow Region. Now, in addition to issues of “Soviet sport”, the service database contains more than 7 million pages of historical documents from the archives of Moscow, Moscow, Irkutsk, Orenburg and Novgorod regions, as well as several municipal archives.

Related posts